Journal of Biogeography (J. Biogeogr.) (2013)

ORIGINAL After continents divide: comparative ARTICLE phylogeography of reef fishes from the and Joseph D. DiBattista1*, Michael L. Berumen2,3, Michelle R. Gaither4, Luiz A. Rocha4, Jeff A. Eble5, J. Howard Choat6, Matthew T. Craig7, Derek J. Skillings1 and Brian W. Bowen1

1Hawai’i Institute of Marine Biology, ABSTRACT Kane’ohe, HI, 96744, USA, 2Red Sea Aim The Red Sea is a biodiversity hotspot characterized by a unique marine Research Center, King Abdullah University of fauna and high endemism. This sea began forming c. 24 million years ago with Science and Technology, Thuwal, Saudi Arabia, 3Biology Department, Woods Hole the separation of the African and Arabian plates, and has been characterized by Oceanographic Institution, Woods Hole, MA, periods of desiccation, hypersalinity and intermittent connection to the Indian 02543, USA, 4Section of Ichthyology, Ocean. We aim to evaluate the impact of these events on the genetic architec- California Academy of Sciences, ture of the Red Sea reef fish fauna. 5 San Francisco, CA, 94118, USA, Department Location Red Sea and Western Indian Ocean. of Biology, University of West Florida, Pensacola, FL, 32514, USA, 6School of Methods We surveyed seven reef fish species from the Red Sea and adjacent Marine and Tropical Biology, James Cook Indian Ocean using mitochondrial DNA cytochrome c oxidase subunit I and University, Townsville, QLD, 4811, Australia, cytochrome b sequences. To assess genetic variation and evolutionary connec- 7Department of Marine Sciences and tivity within and between these regions, we estimated haplotype diversity (h) Environmental Studies, University of San and nucleotide diversity (p), reconstructed phylogenetic relationships among Diego, San Diego, CA, 92110, USA haplotypes, and estimated gene flow and time of population separation using Bayesian coalescent-based methodology. Results Our analyses revealed a range of scenarios from shallow population structure to diagnostic differences that indicate evolutionary partitions and possible cryptic species. Conventional molecular clocks and coalescence analy- ses indicated time-frames for divergence between these bodies of water ranging from 830,000 years to contemporary exchange or recent range expansion. Col- onization routes were bidirectional, with some species moving from the Indian Ocean to the Red Sea compared with expansion out of the Red Sea for other species. Main conclusions We conclude that: (1) at least some Red Sea reef fauna survived multiple salinity crises; (2) endemism is higher in the Red Sea than previously reported; and (3) the Red Sea is an evolutionary incubator, occa- sionally contributing species to the adjacent Indian Ocean. The latter two con- clusions – elevated endemism and species export – indicate a need for enhanced conservation priorities for the Red Sea. *Correspondence: Joseph D. DiBattista, Keywords Hawai’i Institute of Marine Biology, PO Box 1346, Kane’ohe, HI, 96744, USA. Coalescent, cryptic speciation, dispersal, genealogical concordance, gene flow, E-mail: [email protected] mitochondrial DNA, vicariance.

years ago (Ma) (i.e. late-Oligocene period) by the separation INTRODUCTION of the African and Arabian plates, the ocean environment The Red Sea is a deep (maximum depth: 2920 m) and nar- that supports coral reefs originated in the early Pleistocene row (maximum width: 350 km) body of water extending (c. 5 Ma; Siddall et al., 2003; Bosworth et al., 2005). The 2270 km from 30° N in the to 13° N in the Red Sea, which now experiences minimal freshwater inflow . Although the sea began forming c. 24 million and high rates of evaporation, is characterized by

ª 2013 Blackwell Publishing Ltd http://wileyonlinelibrary.com/journal/jbi 1 doi:10.1111/jbi.12068 J. D. DiBattista et al. pronounced north-to-south gradients in salinity (42& to (Randall, 1994), 33% of crustaceans, 15% of echinoderms, 37&), sea surface temperature (winter, 20–28 °C; summer, and up to 25% of corals (Cox & Moore, 2000). Endemism 26–32 °C), and nutrient concentration (low to high) (Raitsos can be even higher for some taxa, reaching 50% in the but- et al., 2011; Ngugi et al., 2012). Oceanographic current pat- terflyfishes (Roberts et al., 1992). terns and climate in the southern Red Sea (and Gulf of Although the evolutionary processes driving high rates of Aden) are heavily influenced by the northern Indian Ocean endemism are unclear, the narrow (18 km) and shallow monsoon system (Smeed, 1997, 2004; Biton et al., 2010), (137 m) strait of the Bab al Mandab, the only connection which reverses wind circulation patterns in the boreal sum- with the Indian Ocean, is likely to have played a key role mer (Southwest Monsoon, April to October) compared to (Klausewitz, 1989). The Red Sea was repeatedly isolated dur- those in the winter (Northeast Monsoon, December to ing Pleistocene glacial cycles when the sea level lowered as March; see Fig. 1). much as 120 m; whether this was achieved through physical Despite its peripheral location relative to the Indo-Pacific, isolation or the restriction of oceanic flow associated with the Red Sea is characterized by high biodiversity, including elevated salinity and temperature remains contentious (Sid- c. 300 reef-building corals (mostly species of and dall et al., 2003; Bailey, 2009). Moreover, cold-water welling Porites; Sheppard & Sheppard, 1991; Riegl & Velimirov, up off Somalia, which precludes reef formation on the 1994) and 1078 fish species (Golani & Bogorodsky, 2010), north-east African and southern Arabian coasts, is likely to which represent key resources for coastal communities. Spe- reinforce this isolation (Smeed, 1997; Kemp, 1998, 2000). cies richness appears to be highest in the central Red Sea, Some authors believe that the Bab al Mandab no longer acts with marked decreases in species abundance and changes in as a physical barrier to dispersal but that an ecological bar- species composition away from this area (DeVantier et al., rier lies within the Red Sea (Ormond & Edwards, 1987). 2000). The Red Sea also harbours one of the highest degrees Roberts et al. (1992) suggested that a turbid-water region of endemism for marine organisms, making up 14% of fishes south of 20° N in the Red Sea may limit larval dispersal, a

(a) (b)

23.5ºN Thuwal (a-g) Al Lith (a-g) Oman (b)

WICC (c) SC

SMC NEC (d) 0º SECC Seychelles (a-f) Diego Garcia (a-g) EACC

(e) SEC

23.5ºS (f) (g)

Sodwana Bay (e)

Figure 1 Scaled map indicating collection sites for all seven reef fish species (a, Acanthurus nigrofuscus;b,Cephalopholis argus;c, Chaetodon auriga;d,Halichoeres hortulanus;e,Lutjanus kasmira;f, sammara;g,Pygoplites diacanthus) sampled in the Red Sea and the Western Indian Ocean. Black lines with arrows show the major current systems flowing through each body of water (abbreviations: EACC, East African Countercurrent; NEC, Northwest Monsoon Current; SC, Somali Current; SEC, South Equatorial Current; SECC, South Equatorial Countercurrent; SMC, Southwest Monsoon Current; WICC, Western Indian Coastal Current). Note the reversing circulation of the SC (from northward to southward), the SMC (from westward to the eastward NEC), the WICC (from eastward to westward), and the current flowing into the Red Sea from the Gulf of Aden (compared with out of the Red Sea and into the Gulf of Aden) during the Northeast Monsoon season (December to March). Site-specific samples sizes are provided in Table 1. (Photo credits: M.L. Berumen, S. Moldzio and L.A. Rocha.)

2 Journal of Biogeography ª 2013 Blackwell Publishing Ltd Phylogeography of Red Sea reef fishes hypothesis which is supported by the presence of a number thereby illuminate evolutionary processes that are the well- of species in the northern/central Red Sea and the Gulf of spring of Red Sea biodiversity. Aden (just outside the Red Sea) that are absent from the southern Red Sea. MATERIALS AND METHODS Despite being acknowledged as a biodiversity hotspot for fishes based on research dating back more than Sample collection 200 years (e.g. Forsskal, 1775), little work has been con- ducted in the Red Sea using modern genetic techniques. Reef fish were collected while or snorkelling Studies in this region tend to focus on the biogeography and at depths of 1–40 m between 2002 and 2011 (Fig. 1, community structure of the more iconic (and endemic) Table 1). Seven reef fish species were targeted: the brown shore fish fauna (e.g. family Chaetodontidae; Righton et al., surgeonfish, Acanthurus nigrofuscus (Forsskal, 1775); the pea- 1996). The majority of genetic studies on reef fishes have cock hind, Cephalopholis argus Schneider, 1801; the threadfin been restricted to the Gulf of Aqaba and northern Red Sea butterflyfish, Chaetodon auriga Forsskal, 1775; the checker- (Hassan et al., 2003; Kochzius & Blohm, 2005; but see board wrasse, Halichoeres hortulanus (Lacepede, 1801); the Froukh & Kochzius, 2007), and few of these considered the bluestripe snapper, Lutjanus kasmira (Forsskal, 1775); the connections between widespread taxa and other biogeograph- Sammara squirrelfish, Neoniphon sammara (Forsskal, 1775); ical provinces, particularly the Indian Ocean (Froukh & and the regal angelfish, Pygoplites diacanthus (Boddaert, Kochzius, 2008). 1772). These species were chosen because they have wide Peripheral reef habitats such as the Red Sea, which forms Indo-Pacific distributions, are abundant, represent a diversity the north-westernmost extension of the Indian Ocean, are of taxonomic families, and can be unequivocally identified typically considered to be biodiversity sinks that receive spe- in the field. Each species was sampled at two locations (Thu- cies from elsewhere but rarely export them (Briggs, 1999). wal and Al Lith) off the coast of the Kingdom of Saudi Ara- The accepted paradigm is therefore that biogeographical bia (KSA) in the central Red Sea, and at one to three sites sinks are ‘evolutionary graveyards’ that do not contribute to in the WIO (oceanic sites: Diego Garcia in the Chagos biodiversity at neighbouring sites. Recent research on reef Archipelago and the Republic of Seychelles; coastal sites: fish and invertebrates, however, demonstrate that peripheral Sodwana Bay, and Al Hallaniyat, Sultanate of regions, such as the Hawaiian Archipelago and the Marque- Oman). Because some of the collections were opportunistic, sas Islands, may act as both a sink and a source, contributing not every species could be sampled at every location (Fig. 1, unique genetic lineages to other regions of the Indo-Pacific Table 1). (Gaither et al., 2010, 2011; DiBattista et al., 2011; Eble et al., 2011; Skillings et al., 2011). Mitochondrial DNA sequencing Our first aim is to assess connections between fauna in the Red Sea and the adjacent Western Indian Ocean Tissue samples were preserved in salt-saturated DMSO (Seu- (WIO) using a molecular genetic approach. The WIO tin et al., 1991). Total genomic DNA was extracted using the forms a biogeographical subdivision of the tropical Indo- ‘HotSHOT’ protocol (Meeker et al., 2007) and subsequently Pacific stretching from East Africa to the Chagos Ridge in stored at À20 °C. Fragments of mtDNA from the cyto- the centre of the Indian Ocean (Sheppard, 2000; Briggs & chrome c oxidase subunit I (COI) and cytochrome b (cyt b) Bowen, 2012). Phylogeographical inferences are strength- genes were amplified using either previously published prim- ened by congruence among multiple species and genes, and ers or modified primers designed for individual species so our study considers seven species of reef fish with wide- (Table 1). These two markers were chosen because they: (1) spread distributions, using two mitochondrial DNA are easy to amplify in most fish; (2) are generally variable at (mtDNA) markers. the population level; (3) facilitate comparisons with pub- Our second aim is to assess whether sea level changes lished sequences; and (4) have had molecular clock rates esti- have influenced extant biodiversity by estimating migration mated based on reef fishes (Bowen et al., 2001; Lessios, 2008; rates and divergence times of reef fishes in the Red Sea and Reece et al., 2010). Polymerase chain reaction (PCR) was WIO. Such analyses will allow us to discriminate between carried out for all species as described in DiBattista et al. the following scenarios: (1) Red Sea populations represent (2012a), with optimal annealing temperatures listed in long-isolated relicts derived from the WIO, which implies Table 1. All samples were sequenced in the forward direction gene flow was absent over the last 5 Myr; (2) Red Sea pop- (and reverse direction for unique or questionable haplotypes) ulations have been isolated from the WIO over evolutionary with fluorescently labelled dye terminators (BigDye version intervals but with recurrent gene flow; or (3) Red Sea pop- 3.1, Applied Biosystems, Foster City, CA, USA) and analysed ulations are the result of recent colonization from the using an ABI 3130xl Genetic Analyzer (Applied Biosystems). WIO, since the Last Glacial Maximum c. 20,000 years ago The sequences were aligned, edited and trimmed to a uni- (Siddall et al., 2003; Bailey, 2009). This dataset provides an form length using Geneious Pro 4.8.4 (Drummond et al., unprecedented opportunity to assess the relationships 2009); unique mtDNA haplotypes were deposited in Gen- between two Indian Ocean biogeographical provinces, and Bank (accession numbers: KC187734–KC188056).

Journal of Biogeography 3 ª 2013 Blackwell Publishing Ltd J. D. DiBattista et al.

Table 1 Study species, number of specimens, fragment length, primers used, and annealing temperatures for mitochondrial DNA cytochrome c oxidase subunit I (COI) and cytochrome b (cyt b) genes. DNA sequences from each primary collection location (Al Lith, Thuwal, Diego Garcia, and the Republic of Seychelles; see text) are described, along with collections made opportunistically at additional locations in the Western Indian Ocean (WIO). All haplotypes are available online in GenBank (accession numbers: KC187734– KC188056).

Molecular sequence data

Red Sea WIO

DNA Fragment Al Diego Other Annealing Species fragment length (bp) Lith Thuwal Garcia Seychelles sites (n) Primer set temp. (°C)

Acanthurus nigrofuscus COI 634 22 27 31 31 – F2–Fish R2 (1) 50 (brown surgeonfish) cyt b 683 22 28 31 30 – Cyb9–Cyb7 (2,3) 58 Cephalopholis argus COI 537 26 19 24 10 Oman (8) Fish F2–Fish R2 (1) 56 (peacock hind) cyt b 618 27 22 32 13 Oman (9) CB6F–CB6R (4) 54 Chaetodon auriga COI 625 27 20 33 30 – Fish F2–Fish R2 (1) 52 (threadfin butterflyfish) cyt b 670 27 20 33 30 – Cyb9–Cyb7 (2,3) 56 Halichoeres hortulanus COI 589 25 27 20 28 – Fish F2–Fish R2 (1) 50 (checkerboard wrasse) cyt b 692 25 27 27 22 – Cyb9–Cyb7 (2,3) 50 Lutjanus kasmira COI 606 21 22 33 20 Sodwana Fish F2–Fish R2 (1) 56 (bluestripe snapper) Bay (34) cyt b 475 23 23 34 19 Sodwana H15020–Cyb5 (5,3) 48 Bay (34) Neoniphon sammara COI 611 20 31 30 28 – Fish F2–Fish R2 (1) 50 (Sammara squirrelfish) cyt b 508 20 31 29 38 – NSAFOR4–NSAREV4* 60 Pygoplites diacanthus COI 634 24 23 33 –– Fish F2–Fish R2 (1) 50 (regal angelfish) cyt b 640 24 23 32 –– PydCytbF3–PydCytbR4* 50

*We designed two novel primer sets to amplify and sequence cyt b for Neoniphon sammara and Pygoplites diacanthus. Their sequences were as follows: NSAFOR4: 5′-TGC CGT GAC GTA AAC TAT GG-3′; NSAREV4: 5′-TGA AGT TGT CGG GAT CTC CT-3′; PydCytbF3: 5′-ATG GCA AAC TTA CGC AAA ACC-3′; PydCytbR4: 5′-GGC TGG TGT GAA GTT GTC-3′. References: (1) Ward et al., 2005; (2) Song et al., 1998; (3) Taberlet et al., 1992; (4) Gaither et al., 2010; (5) Meyer, 1994.

Narum (2006), and negative pairwise Φ values were con- Genetic diversity and population structure ST verted to zeros. To facilitate comparisons among species, an Arlequin 3.5 (Excoffier et al., 2005) was used to calculate additional diversity measure – Jost’s D (Jost, 2008) – was haplotype and nucleotide diversity (h and p, respectively), estimated using spade 1.0 (Chao et al., 2008). and to test for population structure among sampling sites The evolutionary relationship among COI or cyt b haplo- for each species and molecular marker (i.e. 14 total datasets). types was resolved for each species with unrooted networks These analyses were repeated with all Red Sea or WIO sam- constructed with the program network 4.5.1.0 (http://www. ples pooled into two separate regions. Despite the difference fluxus-engineering.com/network_terms.htm) using a median- in the geographical scale of sampling (Red Sea sites, joining algorithm and default settings (as per Bandelt et al., c. 300 km; WIO sites, c. 1000s of km), preliminary work 1999). suggests that the Red Sea haplotypes at Thuwal and Al Lith are consistent with those sampled up to 1200 km north IMA2 analysis (J.D.D., unpublished data), indicating unbiased estimates of genetic diversity within our study range. Because jModel- We calculated the effects of time and gene flow on genetic Test 0.1.1 (Posada, 2008) converged on different models of divergence between populations using Bayesian coalescent- nucleotide sequence evolution among datasets, we calculated based estimation with IMa2 8.26.11 (Hey & Nielsen, 2007; global and pairwise ΦST values based on a HKY model of Hey, 2010). Using F-statistics we determined that samples mutation (Hasegawa et al., 1985). We also ran conventional within regions were not significantly different for all seven frequency-based FST, but the absolute values changed little species after correction for multiple comparisons. We there- and relative values did not change at all; we have therefore fore pooled the Red Sea sites together and the WIO sites elected to report pairwise ΦST. Global ΦST was estimated together, for comparisons between regions for each species using analysis of molecular variance (AMOVA; Excoffier and molecular marker. et al., 1992); deviations from null distributions were tested The isolation-with-migration model implemented in IMa2 with nonparametric permutation procedures (n = 99,999). assumes that two populations of effective size N1 and N2

We controlled for false discovery rate with the method of diverged from an ancestral population (of effective size Na)

4 Journal of Biogeography ª 2013 Blackwell Publishing Ltd Phylogeography of Red Sea reef fishes at time t, and then exchanged migrants at rates m and m . 1.00

1 2 )

h (a) We therefore estimated the time since initial separation or last major colonization event (t), effective population size 0.80

(Ne), and the proportion of migrants arriving into a popula- 0.60 tion per generation (m); all demographic parameters were COI scaled by mutation rate. cyt b 0.40 Mutation rates calibrated in other reef fish based on the closure of the Isthmus of Panama range from 1% to 2% 0.20 per million years for COI and cyt b (Bowen et al., 2001;

Lessios, 2008; Reece et al., 2010). We used a conservative ( diversity Red Sea haplotype 0.00 À estimate of 1.3 9 10 8 mutations per site per year for both 0.00 0.20 0.40 0.60 0.80 1.00 markers (see Lessios, 2008) under a HKY model and a 0.25 WIO haplotype diversity (h) inheritance scalar appropriate for mtDNA. An MCMC chain 0.010 ) (b) with a length of 1,000,000 sampled every 100 generations π with 10% burn-in was used to estimate parameters for each 0.008 species–gene combination. Five independent runs were com- puted to ensure convergence. The independent runs were 0.006 COI subsampled and combined using the ‘L’ mode of IMa2, and cyt b the median values of the parameter distributions for the 0.004 combined runs are presented here. For N. sammara (COI) and P. diacanthus (COI and cyt b), which shared almost no 0.002 haplotypes between regions, prior values of m in both direc- Red Sea nucleotide diversity ( diversity Red Sea nucleotide 0.000 tions were set to zero. Although we regard all absolute 0.000 0.002 0.004 0.006 0.008 0.010 parameter estimates with caution given that our data consist WIO nucleotide diversity (π) of two linked loci, and we apply mutation rates calibrated in other reef fishes, relative comparisons among species are Figure 2 The relationship between (a) haplotype diversity (h) p likely to be robust to such approximations (Karl et al., or (b) nucleotide diversity ( ) estimated for mitochondrial DNA cytochrome c oxidase subunit I (COI; black filled circles) and 2012). cytochrome b (cyt b; open circles) genes in the Red Sea versus Western Indian Ocean (WIO) populations of species considered RESULTS in this study. The black dashed line represents a line of unity, which is the point at which genetic diversity estimates in the Red Sea and WIO are equal within a species. Data points above Genetic diversity and population structure the line of unity indicate greater genetic diversity in the Red Sea, COI and cyt b sequence data revealed divergent patterns of whereas points falling below the line indicate greater genetic diversity in the WIO. genetic diversity and population structure among the seven sampled reef fish species. Haplotype (h) and nucleotide (p) Appendix S2). Estimates of genetic differentiation across all diversity was higher in four out of the seven species in the species were correlated between molecular markers

WIO than in the Red Sea for COI, and in five out of the (nonparametric Spearman’s rank correlation: ΦST, r = 0.77, seven species for cyt b when populations within each region P < 0.001, n = 47; Jost’s D, r = 0.57, P < 0.001, n = 47), were pooled (Fig. 2, and see Appendix S1 in Supporting although the larger spread of Jost’s D values than ΦST values Information). This trend cannot be explained by a greater is probably related to the former not being constrained by sampling effort in the WIO, given that species with compara- within-site heterozygosity. Pairwise genetic differentiation- ble sample sizes for each region, such as Cephalopholis argus based ΦST and Jost’s D were also significantly correlated and P. diacanthus, still had lower genetic diversity in the Red across all datasets (nonparametric Spearman’s rank correla- Sea. tion coefficient: COI,r= 0.67, P < 0.001; cyt b, r = 0.83, AMOVA supported the geographical grouping of sites into P < 0.001).

Red Sea and WIO regions (Table 2). Although there was As expected from the ΦST values, the median-joining some variability in genetic differentiation among sampling networks show more shared COI or cyt b haplotypes sites between regions (Appendix S2), six out of the seven between collection sites within the Red Sea and WIO than species showed significant genetic structure for at least one between these regions (Fig. 4). The only exception to this of the molecular markers (Table 2). pattern was the high proportion of haplotypes shared Population pairwise tests were significantly different in 22 between Al Lith or Thuwal (Red Sea) and Oman (WIO) (for COI) or 16 (for cyt b) out of 47 comparisons (all for Cephalopholis argus. Even though our comparisons P < 0.01); all significant comparisons were between regions between Oman and other sites should be viewed with cau- rather than between sites within regions, and ranged from tion, given that these are based on data from only a single

0.07 to 0.67 for ΦST and 0.05 to 1.00 for Jost’s D (Fig. 3, species (C. argus) with a low sample size (n = 8 or 9),

Journal of Biogeography 5 ª 2013 Blackwell Publishing Ltd J. D. DiBattista et al.

Table 2 Analysis of molecular variance (AMOVA; Excoffier et al., 1992) comparing variation between the Red Sea and Western Indian Ocean populations of reef fish based on mitochondrial DNA cytochrome c oxidase subunit I (COI) and cytochrome b (cyt b) genes. Site-specific samples sizes are shown in Table 1.

Percentage variation

DNA Within Between

Species fragment populations populations Between regions Overall ΦST P Jost’s D (SE)

Acanthurus nigrofuscus COI 76.58 0.25 23.18 0.23 < 0.001 0.59 (0.051) (brown surgeonfish) cyt b 87.96 À0.64 12.68 0.12 < 0.001 0.21 (0.087) Cephalopholis argus COI 72.45 À0.31 27.86 0.28 < 0.001 0.13 (0.088) (peacock hind) cyt b 77.59 1.02 21.39 0.22 < 0.001 0.13 (0.060) Chaetodon auriga COI 80.35 À1.80 21.45 0.20 < 0.001 0.087 (0.041) (threadfin butterflyfish) cyt b 98.48 À0.44 1.96 0.015 0.12 0.021 (0.022) Halichoeres hortulanus COI 97.30 2.60 0.10 0.027 0.071 0.041 (0.019) (checkerboard wrasse) cyt b 95.44 1.08 3.48 0.046 0.045 0.078 (0.088) Lutjanus kasmira COI 100.11 À1.30 1.20 0 0.83 0.006 (0.023) (bluestripe snapper) cyt b 98.88 À0.46 1.58 0.011 0.23 0.090 (0.059) Neoniphon sammara COI 66.03 À0.33 34.30 0.34 < 0.001 0.68 (0.038) (Sammara squirrelfish) cyt b 70.28 À0.69 30.41 0.30 < 0.001 0.59 (0.060) Pygoplites diacanthus COI 30.61 À1.17 70.57 0.69 < 0.001 0.61 (0.015) (regal angelfish) cyt b 65.00 À2.30 37.30 0.35 < 0.001 0.60 (0.027)

) 1.2 IMA2 analysis

b RS versus RS FST t

y RS versus WIO F The estimated times since initial separation between the Red

c ST ( 1.0 Sea and WIO populations for the seven reef fish species ran- n WIO versus WIO FST o

i ged from c. 21,000 to 830,000 years (Table 3). Recent separa- t 0.8 RS versus RS Jost’s D a tions of < 100,000 years were apparent for Chaetodon auriga i t – n RS versus WIO Jost’s D and H. hortulanus, older separations of 100,000 e

r 0.6 300,000 years were apparent for A. nigrofuscus, Cephalopholis e

f WIO versus WIO Jost’s D f

i argus, L. kasmira and N. sammara, and finally P. diacanthus d 0.4 populations have been isolated for 660,000–830,000 years. Of c i t the older separations, L. kasmira was characterized by high e n

e 0.2 subsequent gene flow, whereas gene flow was restricted for g N. sammara (and P. diacanthus); these two species also have e s i the highest level of divergence between the Red Sea and

w 0.0

r Φ

i WIO based on ST (Table 2). Differences among species in a

P both the timing of initial divergence and subsequent migra- -0.2 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 tion rates reveal considerable variation in the link between Pairwise genetic differentiation (COI) Red Sea and WIO populations. The direction of migration varied among species. For Figure 3 The relationship between mitochondrial DNA example, a higher proportion of migrants moved from the cytochrome c oxidase subunit I (COI) and cytochrome b (cyt b) WIO into the Red Sea for H. hortulanus, whereas L. kasmira estimates of pairwise genetic differentiation for the seven species moved predominantly out of the Red Sea (Table 3). For the of reef fish, based on comparisons among Red Sea (RS, remaining species, gene flow was low in both directions, or represented by circles), Western Indian Ocean (WIO; driven by differences in effective population size, indicating represented by triangles), and Red Sea versus WIO sites no bias in effective migration between regions. (represented by squares). Estimates of both ΦST (black symbols) and Jost’s D (grey symbols) are presented here. The black dashed line represents a line of unity, which is the point at DISCUSSION which pairwise genetic differentiation estimates between two study sites are equal for each molecular marker. Data points This study demonstrates barriers to gene flow between the below the line of unity indicate greater genetic differentiation Red Sea and WIO for some reef fish species, but an based on COI, whereas points falling above the line indicate apparent lack of phylogeographical breaks for others, which greater genetic differentiation based on cyt b. may reflect the volatile geological history of the Red Sea some endemic Red Sea fauna do extend to the Omani region. coast (e.g. Cirrhitus spilotoceps, M.R.G. & J.E. Randall, The Red Sea is a marginal water mass whose movement in Bishop Museum, unpublished data). the upper layers is driven by the summer and winter

6 Journal of Biogeography ª 2013 Blackwell Publishing Ltd Phylogeography of Red Sea reef fishes

COI cyt b 20,000 years ago, the inflow pattern and exchange of surface Acanthurus nigrofuscus water was limited, owing to the shallow sill at the Bab al Al Lith Mandab – the only natural gateway into the Red Sea (Siddall Thuwal et al., 2003). As a result, throughout these periods of isola- Sodwana Bay tion, increased evaporation may have raised temperature and Oman salinity levels higher than most reef fishes can tolerate (e.g. Seychelles > 50&; Biton et al., 2008), resulting in periods of reduced Diego Garcia Cephalopholis argus planktonic (i.e. larval) development (Hemleben et al., 1996), and causing mass extirpation within the Red Sea (Sheppard et al., 1992; but see Klausewitz, 1989). In addition to intermittent historical barriers created by Pleistocene glacial cycles, contemporary barriers exist. The Chaetodon auriga lack of coral habitat along the 2200-km coastline from Dji- bouti to southern Somalia may inhibit gene flow between the Red Sea and WIO by limiting opportunities for stepping- stone dispersal (Kemp, 1998). Within the Red Sea, the exten- sive turbid-water region south of 20° N may also inhibit Halichoeres hortulanus larval dispersal or settlement (Ormond & Edwards, 1987; Roberts et al., 1992). The long-term persistence and age of these contemporary barriers, however, is uncertain. Most genetic work on reef fishes within the Red Sea has Lutjanus kasmira focused on the differentiation of fauna between the Gulf of Aqaba and northern Red Sea, with some notable exceptions. Froukh & Kochzius (2008) identified a damselfish in the Red Sea (Chromis viridis) as being distinct from conspecifics in Indonesia and the Philippines based on mtDNA sequences. Similar research on marine invertebrates (Acanthaster planci: Neoniphon sammara Benzie, 1999; Scylla serrata: Fratini & Vannini, 2002) support a genetic distinction of the Red Sea populations. In contrast, Kochzius & Blohm (2005) found no mtDNA differentiation between lionfish (Pterois miles) populations in the Red Sea and Indian Ocean. Five of the seven species we examined were genetically dif- Pygoplites diacanthus ferentiated between the Red Sea and WIO based on AMOVA and median-joining networks. Halichoeres hortulanus and L. kasmira had minimal or inconsistent genetic differentia- tion, as well as extensive mixing of haplotypes within and between regions. Acanthurus nigrofuscus, Cephalopholis argus and Chaetodon auriga had modest differentiation between Figure 4 Median-joining networks showing relationships regions with pronounced separation of peripheral haplotypes, among mitochondrial DNA cytochrome c oxidase subunit I but shared a common haplotype among all sampling sites. (COI) and cytochrome b (cyt b) haplotypes for each study Neoniphon sammara and Pygoplites diacanthus had fixed dif- species (Acanthurus nigrofuscus; Cephalopholis argus; Chaetodon ferences between regions. auriga; Halichoeres hortulanus; Lutjanus kasmira; Neoniphon sammara; Pygoplites diacanthus) collected in the Red Sea (Al Variability in genetic signatures can occur even among Lith and Thuwal) and the Western Indian Ocean (Diego Garcia, closely related species (Rocha et al., 2002; Gaither et al., Oman, Seychelles and Sodwana Bay). Each circle represents a 2010; DiBattista et al., 2012b) and may be related to innate unique haplotype and its size is proportional to its total differences in life history or ecological preferences, although frequency. Branches or black cross-bars represent a single these widely distributed species are all presumably capable of nucleotide change, small black circles represent missing long-distance dispersal (e.g. Eble et al., 2009, 2011; Gaither haplotypes, and colours denote collection location as indicated et al., 2010, 2011) based on available estimates of pelagic lar- by the embedded key. val duration (range: 24–48 days; Thresher & Brothers, 1985; Victor, 1986; Wilson & McCormick, 1999) and longitudinal monsoons acting through a restricted connection with the range size (range: 20,063–21,689 km; Randall, 1999, 2005). adjacent Gulf of Aden (Siddall et al., 2004; Biton et al., Indeed, our study species cover a wide spectrum of dietary 2008). During each glacial maximum of the Pleistocene, the modes ranging from herbivory (A. nigrofuscus) to specialist last characterized by a 120-m drop in sea level only feeding on sessile or mobile invertebrates (Chaetodon auriga,

Journal of Biogeography 7 ª 2013 Blackwell Publishing Ltd J. D. DiBattista et al.

Table 3 Estimates of time in years (t) since initial separation, effective migration rate (2Nem), effective population sizes (Ne), and mutation-scaled migrations rates (m) between Red Sea (RS) and Western Indian Ocean (WIO) populations of seven reef fish species based on mitochondrial DNA cytochrome c oxidase subunit I (COI) and cytochrome b (cyt b) runs in IMa2 (Hey & Nielsen, 2007). Abbreviations: NC, no convergence. Inequalities: posterior probability densities rise to a plateau, so that all estimates larger than the shown value have the same approximate posterior probability.

Effective migration rate Effective population

(2Nem) size (Ne) Migration (m) Initial DNA separation WIO to RS to Species fragment in years (t) WIO to RS RS to WIO WIO RS RS WIO

Acanthurus nigrofuscus (brown surgeonfish) COI 105,000 1.21 0.09 25.10 50.70 0.03 0.01 cyt b 79,000 3.92 0.11 52.75 85.25 0.02 0.001 Cephalopholis argus (peacock bind) COI > 121,000 NC 0.73 0.25 NC 21.05 1.45 cyt b > 212,000 0.01 NC NC 0.60 0.01 6.27 Chaetodon auriga (threadfin butterflyfish) COI 26,800 0.08 0.13 6.60 3.80 0.01 0.01 cyt b 30,700 0.03 4.85 34.65 1.65 0.01 0.07 Halichoeres hortulanus (checkerboard wrasse) COI 26,500 1491.11 0.08 7.50 115.50 6.46 0.01 cyt b 21,600 > 199.50 > 163.50 > 272.50 > 262.50 1.22 0.02 Lutjanus kasmira (bluestripe snapper) COI 41,000 0.23 1282.22 104.50 7.50 0.02 6.14 cyt b 155,000 15.01 104.61 88.10 3.10 2.82 0.50 Neoniphon sammara (Sammara squirrelfish) COI 169,000 0.79 0.03 14.85 20.75 0.02 0.001 cyt b 190,000 0.09 0.24 47.48 17.77 0.0025 0.0025 Pygoplites diacanthus (regal angelfish) COI 831,000 0.01 0.04 4.20 1.40 0.01 0.01 cyt b > 662,000 0.06 0.04 8.88 11.63 0.0025 0.0025

H. hortulanus and P. diacanthus) to piscivory (Cephalopholis (Kemp, 2000). Although we only sampled a few specimens argus, L. kasmira and N. sammara). These species also display (n = 9) of a single species off the coast of Oman (Cephalop- a variety of reproductive behaviours, ranging from dioecism holis argus), these fish were not genetically distinct from con-

(L. kasmira, N. sammara) with mate-pairing (Chaetodon aur- specifics sampled at Diego Garcia (COI: ΦST = 0.029, iga) or spawning aggregations (A. nigrofuscus) to protogyny P = 0.25; cyt b: ΦST < 0.001, P = 0.50) or the Seychelles

(Cephalopholis argus and H. hortulanus). Given that there are (COI: ΦST < 0.001, P = 0.80; cyt b: ΦST = 0.039, P = 0.20). no real unifying life-history features for this diverse group, we suspect that differences in ecological resilience to geological Vicariance events and colonization history disturbance may have contributed to the range of colonization histories, although this will require further testing. Our mtDNA data provide evidence for three separate periods Considering the prevailing currents in the Indian Ocean, it of colonization or export of propagules between the Red Sea is not surprising that sampling sites in the WIO were geneti- and WIO (Table 3). First, Red Sea populations of Chaetodon cally similar to each other. The Chagos and Seychelles archi- auriga and H. hortulanus appear to derive from the WIO pelagos are located in the South Equatorial Current, which during or soon after the most recent glacial maximum flows from east to west. Both archipelagos are also heavily (c. 21,000–31,000 years ago; but see Karl et al., 2012). Sec- influenced by seasonal or permanent countercurrents (South ond, population separations in A. nigrofuscus, Cephalopholis Equatorial Countercurrent and East African Countercurrent, argus and L. kasmira pre-date the Last Glacial Maximum but respectively; Fig. 1). The strong and variable water movement include recurrent gene flow in most cases. Third, N. sammara of the region has resulted in Diego Garcia, which is located at and P. diacanthus represent long-isolated evolutionary lin- the southern end of the Chagos Archipelago, having faunal eages in the Red Sea. These final cases in particular indicate affinities with both the Indo-Polynesian and WIO provinces that some Red Sea residents survived the major temperature (Winterbottom & Anderson, 1997; Craig, 2008; Gaither et al., and salinity crises recorded 19,000, 30,000 and 450,000 years 2011; Briggs & Bowen, 2012). The South African coastline is ago (Siddall et al., 2003). similarly well connected to the central Indian Ocean, being IMa2 analyses indicate bidirectional gene flow between the influenced by the warm Mozambique/Agulhas current (Lut- Red Sea and WIO, which is also apparent in the older his- jeharms, 2006), which facilitates unidirectional (north to tory inscribed in haplotype networks (Fig. 4). Three cases south) transport of tropical fauna from other sites in the WIO. provide especially strong inference: (1) in the COI and cyt b There are several records of long-distance dispersal of network for A. nigrofuscus, the central (ancestral) haplotype tropical reef fish (e.g. Chaetodon zanzibarensis and Ecsenius is observed primarily in the Indian Ocean, whereas the Red lineatus) to the Arabian coastline during periods of upwell- Sea haplotypes are peripheral; (2) in the COI network for N. ing, which indicate that larval transport from the WIO to sammara, the central haplotype is detected only in the Red this region and subsequent settlement are not precluded Sea, with the Indian Ocean haplotypes peripheral; and (3) in

8 Journal of Biogeography ª 2013 Blackwell Publishing Ltd Phylogeography of Red Sea reef fishes the cyt b network for P. diacanthus, the central haplotype is 2012). Here we provide the first multispecies comparison detected only in the Red Sea. Hence the networks for these between Red Sea and Indian Ocean reef fishes, and find a spec- three species indicate colonization into and out of the Red trum of outcomes from recent gene flow to ancient evolution- Sea, which supports the hypothesis that peripheral habitats ary separations. Three broad conclusions are apparent. First, can export biodiversity to the central Indo-Pacific. endemism and biodiversity are higher among Red Sea reef fishes than previously suspected (i.e. N. sammara and P. di- acanthus), and ongoing studies will be likely to elevate these Taxonomic considerations estimates. Second, some elements of the Red Sea fauna sur- Our genetic study highlights three interesting cases where the vived the salinity crises caused by late Pleistocene glaciations. current classification of existing species may not reflect their This does not require continuous residence in the Red Sea, as evolutionary history. Chaetodon auriga is one of the most persistence in the Gulf of Aden just outside the Red Sea widespread butterflyfishes on the planet, with a distribution remains a possibility. It seems unlikely, however, that a geneti- of c. 82.2 million km2 across the tropical Indo-Pacific (Allen cally distinct and cohesive fauna could survive in the Gulf et al., 1998). The original species description is from Red Sea without extensive admixture with other Indian Ocean popula- specimens, which lack a dark spot on the margin of the soft tions. We therefore favour the explanation that Red Sea refugia dorsal fin, such that conspecifics outside the Red Sea were existed during low sea level stands associated with glaciations. regarded as the subspecies C. auriga setifer (Bloch, 1795). Third, peripheral habitats such as marginal seas and isolated Although we did detect differences in mtDNA sequences archipelagos are not necessarily ‘evolutionary graveyards’. between the Red Sea and WIO, these were only marginally Rather, our data indicate that the Red Sea is capable of export- significant. In addition, the most common haplotype was ing biodiversity to the broader Indo-Pacific, thus operating as shared between the Red Sea and WIO at both COI and cyt a potential engine of evolutionary diversity in our oceans. b. Even though colour morphs do correspond to genetic par- titions in some species (e.g. Craig & Randall, 2008; Drew ACKNOWLEDGEMENTS et al., 2008; Randall & Rocha, 2009), discordance between genetic divergence and coloration is well documented in reef This research was supported by National Science Foundation fishes (Ramon et al., 2003; Rocha, 2004; Messmer et al., grants OCE-0453167 and OCE-0929031 to B.W.B., National 2005), including butterflyfishes (family Chaetodontidae: Geographic Society Grant 9024-11 to J.D.D., KAUST Red McMillan & Palumbi, 1995) and their sister group, the Sea Research Center funding to M.L.B., California Academy angelfishes (family Pomacanthidae: Bowen et al., 2006; of Sciences funding to L.A.R., and by a Natural Sciences and DiBattista et al., 2012a). For these reasons, we regard the Red Engineering Research Council of Canada (NSERC) postgrad- Sea population as conspecific with all other populations of uate fellowship to J.D.D. For specimen collections, we thank Chaetodon auriga, although the shallow but significant popu- Gavin Gouws (South Africa Institute for Aquatic Biodiver- lation genetic differentiation supports the subspecific status. sity), Matthew Iacchei, Kelton W. McMahon, Gerrit Nan- Two species in our study reveal a strikingly different pat- ninga, Jonathan Puritz and Charles R.C. Sheppard. We also tern: N. sammara and P. diacanthus were characterized by thank Robert J. Toonen, Serge Planes, John E. Randall, Clau- high ΦST (or Jost’s D) values relative to all other species and dia Rocha, Jo-Ann C. Leong, Eric Mason at Dream Divers, strong mtDNA differences between regions. For P. diacanthus David Pence, the KAUST Coastal and Marine Resources in particular, Red Sea and WIO haplotypes are separated by Core Lab, the Administration of the British Indian Ocean at least three mutations at cyt b and one fixed mutation at Territory, and members of the ToBo lab for logistic support; COI, indicating isolation for several hundred thousand years. we thank Stephan Moldzio for photos of Neoniphon sammara; This genetic separation is matched by coloration differences we thank the Center for Genomics, Proteomics, and Bioinfor- between Red Sea and WIO populations (Allen et al., 1998), matics at the University of Hawai’i at Manoa, in addition to indicating long-isolated populations that, unlike other exam- the KAUST Bioscience Core Facility for their assistance with ined species, failed to reconnect during interglacial periods. DNA sequencing. This is contribution no. 1530 from the Notably, the species is absent from sites in the Arabian Sea, Hawai’i Institute of Marine Biology and no. 8790 from the indicating geographical isolation (Kemp, 1998). While School of Ocean and Earth Science and Technology. we know of no coloration or morphological differences in N. sammara that may indicate cryptic lineages, this possibil- REFERENCES ity merits further investigation. Allen, G.R., Steene, R. & Allen, M. (1998) A guide to angelfishes and butterflyfishes. Odyssey Publishing, Perth. CONCLUSIONS Avise, J.C. (1992) Molecular population structure and the Comparative phylogeographical studies have done much to biogeographic history of a regional fauna: a case history illuminate the evolutionary history of regional marine faunas with lessons for conservation biology. Oikos, 3,62–76. (Avise, 1992; Lessios & Robertson, 2006; Kelly & Palumbi, Bailey, G. (2009) The Red Sea, coastal landscapes, and homi- 2010; Carpenter et al., 2011; Toonen et al., 2011; Ludt et al., nin dispersals. The evolution of human populations in Ara-

Journal of Biogeography 9 ª 2013 Blackwell Publishing Ltd J. D. DiBattista et al.

bia (ed. by M. Petraglia and J. Rose), pp. 15–37. Springer, nectivity and a cryptic endemic species in the Hawaiian Dordrecht. archipelago. Journal of Marine Biology, 2011, 839134. Bandelt, H.J., Forster, P. & Rohl,€ A. (1999) Median-joining DiBattista, J.D., Waldrop, E., Bowen, B.W., Schultz, J.K., networks for inferring intraspecific phylogenies. Molecular Gaither, M.R., Pyle, R.L. & Rocha, L.A. (2012a) Twisted Biology and Evolution, 16,37–48. sister species of pygmy angelfishes: discordance between Benzie, J.A.H. (1999) Major genetic differences between , coloration, and phylogenetics. Coral Reefs, 31, crown-of-thorns starfish (Acanthaster planci) populations 839–851. in the Indian and Pacific Oceans. Evolution, 53, 1782–1795. DiBattista, J.D., Rocha, L.A., Craig, M.T., Feldheim, K.A. & Biton, E., Gildor, H. & Peltier, W.R. (2008) Red Sea during Bowen, B.W. (2012b) Phylogeography of two closely the Last Glacial Maximum: implications for sea level related Indo-Pacific butterflyfishes reveals divergent evolu- reconstruction. Paleoceanography, 23, PA1214. tionary histories and discordant results from mtDNA and Biton, E., Gildor, H., Trommer, G., Siccha, M., Kucera, M., microsatellites. Journal of Heredity, 103, 617–629. van der Meer, M.T.J. & Schouten, S. (2010) Sensitivity of Drew, J., Allen, G.R., Kaufman, L. & Barber, P.H. (2008) Red Sea circulation to monsoonal variability during the Endemism and regional color and genetic differences in Holocene: an integrated data and modeling study. Paleoce- five putatively cosmopolitan reef fishes. Conservation Biol- anography, 25, PA4209. ogy, 22, 965–975. Bosworth, W., Huchon, P. & McClay, K. (2005) The Red Sea Drummond, A.J., Ashton, B., Cheung, M., Heled, J., Kearse, and Gulf of Aden Basins. Journal of African Earth Sciences, M., Moir, R., Stones-Havas, S., Thierer, T. & Wilson, A. 43, 334–378. (2009) Geneious v4.8. Available at: http://www.geneious. Bowen, B.W., Bass, A.L., Rocha, L.A., Grant, W.S. & Robert- com/. son, D.R. (2001) Phylogeography of the trumpetfishes Eble, J.A., Toonen, R.J. & Bowen, B.W. (2009) Endemism (Aulostomus): ring species complex on a global scale. Evo- and dispersal: comparative phylogeography of three sur- lution, 55, 1029–1039. geonfishes across the Hawaiian Archipelago. Marine Biol- Bowen, B.W., Muss, A., Rocha, L.A. & Grant, W.S. (2006) ogy, 156, 689–698. Shallow mtDNA coalescence in Atlantic pygmy angelfishes Eble, J.A., Toonen, R.J., Sorenson, L., Basch, L.V., Papastam- (genus Centropyge) indicates a recent invasion from the atiou, Y.P. & Bowen, B.W. (2011) Escaping paradise: larval Indian Ocean. Journal of Heredity, 97,1–12. export from Hawaii in an Indo-Pacific reef fish, the yellow Briggs, J.C. (1999) Coincident biogeographic patterns: Indo- tang (Zebrasoma flavescens). Marine Ecology Progress Series, West Pacific Ocean. Evolution, 53, 326–335. 428, 245–258. Briggs, J.C. & Bowen, B.W. (2012) A realignment of marine Excoffier, L., Smouse, P.E. & Quattro, J.M. (1992) Analysis biogeographic provinces with particular reference to fish of molecular variance inferred from metric distances distributions. Journal of Biogeography, 39,12–30. among DNA haplotypes: application to human mitochon- Carpenter, K.E., Barber, P.H., Crandall, E.D., Ablan-Lagman, drial DNA restriction data. Genetics, 131, 479–491. M.C.A., Ambariyanto, Mahardika, G.N., Manjaji-Matsum- Excoffier, L., Laval, G. & Schneider, S. (2005) Arlequin version oto, B.M., Juinio-Menez,~ M.A., Santos, M.D., Starger, C.J. 3.0: an integrated software package for population genetics & Toha, A.H.A. (2011) Comparative phylogeography of data analysis. Evolutionary Bioinformatics Online, 1,47–50. the Coral Triangle and implications for marine manage- Forsskal, P. (1775) Descriptiones animalium avium, amphibio- ment. Journal of Marine Biology, 2011, 396982. rum, piscium, insectorum, vermium; quae in itinere orientali Chao, A., Jost, L., Chiang, S.C., Jiang, Y.-H. & Chazdon, R.L. observavit.Moller,€ Copenhagen, Denmark. (2008) A two stage probabilistic approach to multiple- Fratini, S. & Vannini, M. (2002) Genetic differentiation in community similarity indices. Biometrics, 64, 1178–1186. the mud Scylla serrata (Decapoda: Portunidae) within Cox, C.B. & Moore, P.D. (2000) Biogeography: an ecological the Indian Ocean. Journal of Experimental Marine Biology and evolutionary approach, 6th edn. Blackwell, Oxford. and Ecology, 272, 103–116. Craig, M.T. (2008) The goldrim surgeonfish (Acanthurus nig- Froukh, T. & Kochzius, M. (2007) Genetic population struc- ricans; Acanthuridae) from Diego Garcia, Chagos Archipel- ture of the endemic fourline wrasse (Larabicus quadrilinea- ago: first record for the central Indian Ocean. Zootaxa, tus) suggests limited larval dispersal distances in the Red 1850,65–68. Sea. Molecular Ecology, 16, 1359–1367. Craig, M.T. & Randall, J.E. (2008) Two new species of the Froukh, T. & Kochzius, M. (2008) Species boundaries and Indo-Pacific clingfish genus Discotrema (Gobiesocidae). evolutionary lineages in the blue green damselfishes Chr- Copeia, 2008,68–74. omis viridis and Chromis atripectoralis (Pomacentridae). DeVantier, L.M., Turak, E., Al-Shaikh, K.A. & De’ath, G. Journal of Fish Biology, 72, 451–457. (2000) Coral communities of the central-northern Saudi Gaither, M.R., Toonen, R.J., Robertson, D.R., Planes, S. & Arabian Red Sea. Fauna of Arabia, 18,23–66. Bowen, B.W. (2010) Genetic evaluation of marine biogeo- DiBattista, J.D., Wilcox, C., Craig, M.T., Rocha, L.A. & Bo- graphical barriers: perspectives from two widespread Indo- wen, B.W. (2011) Phylogeography of the Pacific Blueline Pacific snappers (Lutjanus kasmira and Lutjanus fulvus). Surgeonfish, Acanthurus nigroris, reveals high genetic con- Journal of Biogeography, 37, 133–147.

10 Journal of Biogeography ª 2013 Blackwell Publishing Ltd Phylogeography of Red Sea reef fishes

Gaither, M.R., Bowen, B.W., Bordenave, T.-R., Rocha, L.A., Ludt, W.B., Bernal, M., Bowen, B.W. & Rocha, L.A. (2012) Newman, S.J., Gomez, J.A., van Herwerden, L. & Craig, Living in the past: phylogeography and population histories M.T. (2011) Phylogeography of the reef fish Cephalopholis of Indo-Pacific wrasses (genus Halichoeres) in shallow argus (Epinephelidae) indicates Pleistocene isolation across lagoons versus outer reef slopes. PLoS ONE, 7, e38042. the Indo-Pacific barrier with contemporary overlap in the Lutjeharms, J.R.E. (2006) The Agulhas Current. Springer, Coral Triangle. BMC Evolutionary Biology, 11, 189. New York, NY. Golani, D. & Bogorodsky, S.V. (2010) The fishes of the Red McMillan, W.O. & Palumbi, S.R. (1995) Concordant evolution- Sea – reappraisal and updated checklist. Zootaxa, 2463,1– ary patterns among Indo-West Pacific butterflyfishes. Proceed- 135. ings of the Royal Society B: Biological Sciences, 260,229–236. Hasegawa, M., Kishino, H. & Yano, T.-A. (1985) Dating of Meeker, N.D., Hutchinson, S.A., Ho, L. & Trede, N.S. (2007) the human-ape splitting by a molecular clock of mitochon- Method for isolation of PCR-ready genomic DNA from drial DNA. Journal of Molecular Evolution, 22, 160–174. zebrafish tissues. BioTechniques, 43, 610–614. Hassan, M., Harmelin-Vivien, M. & Bonhomme, F. (2003) Messmer, V., van Herwerden, L., Munday, P.L. & Jones, G.P. Lessepsian invasion without bottleneck: example of two (2005) Phylogeography of colour polymorphism in the rabbitfish species (Siganus rivulatus and S. luridus). Journal coral reef fish Pseudochromis fuscus, from Papua New Gui- of Experimental Marine Biology and Ecology, 291, 219–232. nea and the Great Barrier Reef. Coral Reefs, 24, 392–402. Hemleben, C., Meishchner, D., Zahn, R., Almogi-Labin, A., Meyer, A. (1994) Shortcomings of the cytochrome b gene as Erlenkeuser, H. & Hiller, B. (1996) Three hundred eighty a molecular marker. Trends in Ecology and Evolution, 9, thousand year long stable isotope and faunal records from 278–280. the Red Sea: influence of global sea level change on Narum, S.R. (2006) Beyond Bonferroni: less conservative hydrography. Paleoceanography, 11, 147–156. analyses for conservation genetics. Conservation Genetics, Hey, J. (2010) Isolation with migration models for more 7, 783–787. than two populations. Molecular Biology and Evolution, 27, Ngugi, D.K., Antunes, A., Brune, A. & Stingl, U. (2012) Bio- 905–920. geography of pelagic bacterioplankton across an antagonis- Hey, J. & Nielsen, R. (2007) Integration within the Felsen- tic temperature–salinity gradient in the Red Sea. Molecular stein equation for improved Markov chain Monte Carlo Ecology, 21, 388–405. methods in population genetics. Proceedings of the Ormond, R. & Edwards, A. (1987) Red Sea fishes. Key envi- National Academy of Sciences USA, 104, 2785–2790. ronments – Red Sea (ed. by A.J. Edwards and S.M. Head),

Jost, L. (2008) GST and its relatives do not measure differen- pp. 251–287. Pergamon Press, Oxford. tiation. Molecular Ecology, 17, 4015–4026. Posada, D. (2008) jModelTest: phylogenetic model averaging. Karl, S.A., Toonen, R.J., Grant, W.S. & Bowen, B.W. (2012) Molecular Biology and Evolution, 25, 1253–1256. Common misconceptions in molecular ecology: echoes of Raitsos, D.E., Hoteit, I., Prihartato, P.K., Chronis, T., Trian- the modern synthesis. Molecular Ecology, 21, 4171–4189. tafyllou, G. & Abualnaja, Y. (2011) Abrupt warming of the Kelly, R.P. & Palumbi, S.R. (2010) Genetic structure among Red Sea. Geophysical Research Letters, 38, L14601. 50 species of the northeastern Pacific rocky intertidal com- Ramon, M.L., Lobel, P.S. & Sorenson, M.D. (2003) Lack of munity. PLoS ONE, 5, e8594. mitochondrial genetic structure in hamlets (Hypoplectrus Kemp, J. (1998) Zoogeography of the coral reef fishes of the spp.): recent speciation or ongoing hybridization? Molecu- Socotra Archipelago. Journal of Biogeography, 25, 919–933. lar Ecology, 12, 2975–2980. Kemp, J. (2000) Zoogeography of the coral reef fishes of the Randall, J.E. (1994) Twenty-two new records of fishes from northeastern Gulf of Aden, with eight new records of coral the Red Sea. Fauna of Saudi Arabia, 14, 259–275. reef fishes from Arabia. Fauna of Arabia, 18, 293–321. Randall, J.E. (1999) Report on fish collections from the Pit- Klausewitz, W. (1989) Evolutionary history and zoogeogra- cairn Islands. Atoll Research Bulletin, 461,1–53. phy of the Red Sea ichthyofauna. Fauna of Saudi Arabia, Randall, J.E. (2005) Reef and shore fishes of the South Pacific: 10, 310–337. New Caledonia to Tahiti and the Pitcairn Islands. Univer- Kochzius, M. & Blohm, D. (2005) Genetic population struc- sity of Hawai’i Press, Honolulu, HI. ture of the lionfish Pterois miles (Scorpaenidae, Pteroinae) Randall, J.E. & Rocha, L.A. (2009) Chaetodontoplus poliourus,a in the Gulf of Aqaba and northern Red Sea. Gene, 347, new angelfish (Perciformes: Pomacanthidae) from the tropical 295–301. western Pacific. The Raffles Bulletin of Zoology, 57,511–520. Lessios, H.A. (2008) The Great American Schism: divergence Reece, J.S., Bowen, B.W., Smith, D.G. & Larson, A. (2010) of marine organisms after the rise of the Central American Molecular phylogenetics of moray eels (Muraenidae) dem- isthmus. Annual Review of Ecology, Evolution, and System- onstrates multiple origins of a shell-crushing jaw (Gymnom- atics, 39,63–91. uraena, Echidna) and multiple colonizations of the Atlantic Lessios, H.A. & Robertson, D.R. (2006) Crossing the impassi- Ocean. Molecular Phylogenetics and Evolution, 57, 829–835. ble: genetic connections in 20 reef fishes across the eastern Riegl, B. & Velimirov, B. (1994) The structure of coral com- Pacific barrier. Proceedings of the Royal Society B: Biological munities at Hurghada in the Northern Red Sea. Marine Sciences, 273, 2201–2208. Ecology, 15, 213–231.

Journal of Biogeography 11 ª 2013 Blackwell Publishing Ltd J. D. DiBattista et al.

Righton, D., Kemp, J. & Ormond, R. (1996) Biogeography, Toonen, R.J., Andrews, K.R., Baums, I.B., Bird, C.E., Con- community structure and diversity of Red Sea and western cepcion, G.T., Daly-Engel, T.S., Eble, J.A., Faucci, A., Gai- Indian Ocean butterflyfishes. Journal of the Marine Biologi- ther, M.R., Iacchei, M., Puritz, J.B., Schultz, J.K., Skillings, cal Association of the UK, 76, 223–228. D.J., Timmers, M.A. & Bowen, B.W. (2011) Defining Roberts, C.M., Shepherd, A.R.D. & Ormond, R.F.G. (1992) boundaries for ecosystem-based management: a multispe- Large-scale variation in assemblage structure of Red Sea cies case study of marine connectivity across the Hawaiian butterflyfishes and angelfishes. Journal of Biogeography, 19, archipelago. Journal of Marine Biology, 2011, 460173. 239–250. Victor, B.C. (1986) Duration of the planktonic larval stage of Rocha, L.A. (2004) Mitochondrial DNA and color pattern var- one hundred species of Pacific and Atlantic wrasses (family iation in three western Atlantic Halichoeres (Labridae), with Labridae). Marine Biology, 90, 317–326. the revalidation of two species. Copeia, 2004, 770–782. Ward, R.D., Zemlak, T.S., Innes, B.H., Last, P.R. & Hebert, Rocha, L.A., Bass, A.L., Robertson, D.R. & Bowen, B.W. P.D.N. (2005) DNA barcoding Australia’s fish species. (2002) Adult habitat preferences, larval dispersal, and the Philosophical Transactions of the Royal Society B: Biological comparative phylogeography of three Atlantic surgeonfish- Sciences, 360, 1847–1857. es (Teleostei: Acanthuridae). Molecular Ecology, 11, 243– Wilson, D.T. & McCormick, M.I. (1999) Microstructure of 251. settlement-marks in the otoliths of tropical reef fish. Mar- Seutin, G., White, B.N. & Boag, P.T. (1991) Preservation of ine Biology, 134,29–41. avian blood and tissue samples for DNA analyses. Cana- Winterbottom, R. & Anderson, R.C. (1997) A revised check- dian Journal of Zoology, 69,82–90. list of the epipelagic and shore fishes of the Chagos Archi- Sheppard, C.R.C. (2000) Coral reefs of the western Indian pelago, Central Indian Ocean. Ichthyological Bulletin, 66, Ocean: an overview. Coral reefs of the Indian Ocean: their 1–28. ecology and conservation (ed. by T.R. McClanahan, C.R.C. Sheppard and D.O. Obura), pp. 3–38. Oxford University Press, Oxford. Sheppard, C.R.C. & Sheppard, A.L.S. (1991) Corals and coral SUPPORTING INFORMATION 12 – communities of Arabia. Fauna of Saudi Arabia, ,3 170. Additional Supporting Information may be found in the Sheppard, C.R.C., Price, A.R.G. & Roberts, C.M. (1992) Mar- online version of this article: ine ecology of the Arabian region: patterns and processes in extreme tropical environments. Academic Press, London. Appendix S1 Mitochondrial DNA haplotype (h) and nucle- p Siddall, M., Rohling, E.J., Almogi-Labin, A., Hemleben, C., otide diversity ( ) for each species from each collection loca- Meischner, D., Schmelzer, I. & Smeed, D.A. (2003) Sea- tion. Φ level fluctuations during the last glacial cycle. Nature, 423, Appendix S2 Population pairwise ST and Jost’s D values 853–858. for each species based on mitochondrial DNA sequences. Siddall, M., Smeed, D.A., Hemleben, C., Rohling, E.J., Sch- melzer, I. & Peltier, W.R. (2004) Understanding the Red Sea response to sea level. Earth and Planetary Science Let- BIOSKETCH ters, 225, 421–434. Skillings, D.J., Bird, C.E. & Toonen, R.J. (2011) Gateways to The authors’ interests are focused on illuminating the evolu- Hawai’i: genetic population structure of the tropical sea tionary processes that generate marine biodiversity. They cucumber Holothuria atra. Journal of Marine Biology, 2011, have carried out phylogeographical surveys of over 20 reef 783030. fish species in the greater Indo-Pacific to test existing evolu- Smeed, D. (2004) Exchange through the Bab el Mandab. tionary models, resolve the life-history traits that influence Deep-Sea Research Part II: Tropical Studies in Oceanogra- dispersal and population separations in reef organisms, and phy, 51, 455–474. inform marine conservation (e.g. defining the boundaries of Smeed, D.A. (1997) Seasonal variation of the flow in the marine protected areas). 20 – strait of Bab al Mandab. Oceanologica Acta, , 773 781. Author contributions: J.D.D. conceived the ideas for this Song, C.B., Near, T.J. & Page, L.M. (1998) Phylogenetic rela- study, collected tissue samples and produced DNA sequences, tions among percid fishes as inferred from mitochondrial analysed the data, and led the writing. In addition to con- cytochrome b DNA sequence data. Molecular Phylogenetics tributing to writing, M.L.B., B.W.B., J.H.C. and M.T.C. col- 10 – and Evolution, , 343 353. lected tissue samples, M.R.G. and L.A.R. collected tissue Taberlet, P., Meyer, A. & Bouvet, J. (1992) Unusual mito- samples and produced DNA sequences, J.A.E. produced chondrial variation in two local populations of blue tit DNA sequences, and D.J.S. implemented and interpreted 1 – Parus caeruleus. Molecular Ecology, ,27 36. coalescent analyses. Thresher, R.E. & Brothers, E.B. (1985) Reproductive ecology and biogeography of Indo-West Pacific angelfish (Pisces: Pomacanthidae). Evolution, 39, 878–887. Editor: Craig McClain

12 Journal of Biogeography ª 2013 Blackwell Publishing Ltd